Found a couple of interesting pages about this rare machine:
Ram’s Totally Unoffical ATW800 Homepage
and
It’s not much to look at - image above from the Crummy Computers Wiki - but of course it’s what’s on the inside that counts, at least in this case.
Here’s a repair video (German audio)
(Ram Meenakshisundaram’s Transputer pages are well worth exploring)
following on from the discussion at ST Book, the Notebook Atari ST:
I remember Atari holding an ATW seminar at University of Strathclyde’s Mechanical Engineering department when I was an undergrad. I may even have seen an ATW demonstrated!
The department had bought a couple of labs full of STs a few years before, and all undergrads learnt basic programming on them. We used Computer Concepts’ Fast BASIC, a very nice dialect of BBC BASIC with a full GEM IDE that ran from the cartridge slot. Maybe Atari UK had hoped that this would evolve into some kind of subsidized deal for students like had happened with Sinclair QLs a few years before.
Transputer machines were touted as being the next big thing for finite element analysis. Whether they were actually good for that (outside the huge machines made by Meiko) I don’t know.
I’m pretty sure that nothing happened with Atari. Silicon Graphics machines started to appear, and by the time I’d finished my postgrad studies, the old computer lab was full of SGIs.
I was at the Comdex show in Vegas when Atari introduced what was originally called the Abaq. I was there representing Commodore/Amiga, of course, and we used to regularly go to the Atari booth to point and laugh. But this one has seemed kind of cool, full of potential. They had hooked up with Tim King, the guy who did the AmigaDOS subsystem for Amigas back in 1984, and while he always seemed to have some weird ideas, that wasn’t necessarily a bad thing.
I had, of course, learned all about the Transputer and in particular, the idea that INMOS wanted “transputers” to be kind of ubiquitious – you’d build one in to anything you wanted to control: displays, storage, etc. and make them all smart and also connected via standard multiprocessing protocols to a main computer that might have its own Transputer cluster. The Abaq had parts of an Atari ST acting as a front end to whatever the included T800, and presumably any additional Transputers you connected up, were doing.
Curiously, today we have a wide variety of computing elements, but in many places, graphics, storage, etc. are hooked up via [one or more] PCIe links. So INMOS was one of the first to really get where 21st Century computing was going. At Commodore, I had already embraced that idea… that’s also how the Commodore VIC-20, Commodore 64, and Commodore 128 (which I worked on) did their I/O – though a serial link to smart peripherals. Yeah, it was slow as balls, but in many ways the right idea, just held back by the 8-bit tech of the day… and a bug or two in the original 6522 GPIO chips used in the VIC-20.
Rumour within what used to be INMOS is that IEEE 1394 serial - aka Firewire - picked up the trick of Data/Strobe encoding from the new inter-transputer links introduced with the C104 and T9000 family. The older links were dubbed OSLinks and the new ones DSLinks.
Just found this video of an ATW rendering a couple images, using 4 processors. This video is sped up 4x. The person who posted it said in real time, it takes 8 mins. He also said the “salt” molecule rendering that you see at the beginning took 45 mins.
It was also used in the IEEE 1355 standard and so in the SpaceWire standard.
aha: " Historically, IEEE 1355 derived from the asynchronous serial networks developed for the Transputer model T9000 on-chip serial data interfaces"
The ABAQ also introduced Atari to Richard Miller, who later followed Shiraz Shivji in the VP Hardware role.
